Developing and Validating a Nationwide Risk-Stratification Prediction Model for Post-General Anesthetic Mortality-The Potentially Protective Effects of Anti-Lipid, Hypoglycaemic, and Anti-Hypertensive Agents in Taiwanese Geriatric Group.

Background: This study used the National Taiwan Health Insurance database, including prescribed medications, underlying systemic diseases, and surgical types, to evaluate the relationship between those factors and mortality after general anesthesia.

Method: This retrospective, population-based study was retrieved from the National Health Insurance Research Database in Taiwan between January 1, 2005, and December 31, 2013. We proposed a logistic regression model that can significantly predict postanesthesia mortality, which was validated in another set of databases. Moreover, we subgroup the study population into four major comparison groups, which include major organ transplantation, cardiovascular (CV) surgery, major neurosurgery, and others according to the diagnostic codes of the International Classification of Diseases, ninth revision, clinical modification codes.

Results: A total of 681,779 subjects were analyzed in this study, and the most common comorbidity was hypertension. Age was an important determinant associated with postoperative mortality among different surgical types. Perioperative prescription could reduce the risks of the operation. The major surgeries (major organ transplantation, CV surgery, and neurosurgery) were associated with increasing postoperative mortality. The prediction model performed well in the validation group (area under receiver operating characteristic curve [AUROC] = 0.8753 for in-hospital mortality; AUROC = 0.8767 for 30-day mortality).

Conclusions: The prediction model that we proposed from this nationwide study could identify the predictors for postoperative mortality. The potentially protective effects of anti-lipid, hypoglycemic, and anti-hypertensive agents were encouraging in the geriatric preoperative group. It is expected that applying this prediction model and prescription into clinical practice could improve surgical risk stratification and further improve patient outcomes.